Analytica Chimica Acta 581 (2007) 193–201

Towards multi-colour strategies for the detection of oligonucleotide

hybridization using quantum dots as energy donors in
fluorescence resonance energy transfer (FRET)
W. Russ Algar, Ulrich J. Krull ∗
Chemical Sensors Group, Department of Chemical and Physical Sciences, University of Toronto at Mississauga,
3359 Mississauga Road North, Mississauga, Ontario L5L 1C6, Canada
Received 26 June 2006; received in revised form 14 August 2006; accepted 14 August 2006
Available online 22 August 2006

Abstract
The potential for a simultaneous two-colour diagnostic scheme for nucleic acids operating on the basis of fluorescence resonance energy
transfer (FRET) has been demonstrated. Upon ultraviolet excitation, two-colours of CdSe/ZnS quantum dots with conjugated oligonucleotide
probes act as energy donors yielding FRET-sensitized acceptor emission upon hybridization with fluorophore (Cy3 and Alexa647) labeled target
oligonucleotides. Energy transfer efficiencies, Förster distances, changes in quantum yield and lifetime, and signal-to-noise with respect to non-
specific adsorption have been investigated. The dynamic range and limit-of-detection are tunable with the concentration of QD–DNA conjugate.
The Cy3 and Alexa647 acceptor schemes can detect target from 4 to 100% or 10 to 100% of the QD–DNA conjugate concentration, respectively.
Nanomolar limits of detection have been demonstrated in this paper, however, results indicate that picomolar detection limits can be achieved with
standard instrumentation. The use of an intercalating dye (ethidium bromide) as an acceptor to alleviate non-specific adsorption is also described
and increases signal-to-noise from S/N < 2 to S/N = 9–10. The ethidium bromide system had a dynamic range from 8 to 100% of the QD–DNA
conjugate concentration and could detect target in a matrix containing an excess of non-complementary nucleic acid.
© 2006 Elsevier B.V. All rights reserved.

Keywords: Quantum dots; Energy transfer; Deoxyribonucleic acid; Ethidium bromide; Biosensor

1. Introduction
Quantum dots (QDs), or colloidal semiconductor nanocrys-
tals, have been the center of much attention in the past few years.
The unique photophysical properties of these particles, particu-
larly in terms of brightness, photostability, narrow tunable emis-
sion, and broad absorption, are attractive in many applications.
One area of application has been imaging with quantum dots
[1–6], where the optical properties are advantageous in com-
parison to most organic fluorophores. Another growing area of
application has been diagnostics. For example, antibodies have
been conjugated to quantum dots for use in sandwich immunoas-
says [7], and in fluorescence resonance energy transfer (FRET)
based sensors for 2,4,6-trinitrotoluene (TNT) [8]. In other FRET
motifs, quantum dots have been conjugated to DNA hairpins for
∗ Corresponding author. Tel.: +1 905 828 5437; fax: +1 905 828 5425.
E-mail address: ukrull@utm.utoronto.ca (U.J. Krull).
use in molecular beacons [9], have been used to detect hybridiza-
tion events [10–12], and have also been conjugated to peptides
and proteins for detection of enzymes [13] and small molecules
[14]. We report an initial investigation of the development of a
FRET and QD-based strategy for the detection of nucleic acids,
with the long term goal of achieving a multi-colour diagnostic
technology suitable for the simultaneous detection of multiple
sequences.
In general, most multi-colour optical diagnostic or imaging
schemes require multiple excitation sources. One of the more
popular optical diagnostic schemes for nucleic acid detection is
fiber-optic biosensors, which have been developed for the detec-
tion of pathogens [15,16] and genetic analyses [17,18]. However,
the simultaneous detection of multiple sequences requires either
two-colour excitation or discrete sensor elements exposed to
a common analyte solution. Other conventional optical tech-
nologies for multiplexed analysis, including both fluorescence
and surface plasmon resonance (SPR) imaging of nucleic acid
microarrays, require fabrication of discrete sensing elements

0003-2670/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.aca.2006.08.026
194 W.R. Algar, U.J. Krull / Analytica Chimica Acta 581 (2007) 193–201

Fig. 1. Proposed QD–FRET-based strategy for two-colour nucleic acid detection. (a) Simultaneous and efficient excitation of green and red quantum dots was possible
in the ultraviolet-region without significant excitation of Cy3 or Alexa647 in solution. When probe oligonucleotides were conjugated to QDs, hybridization with a
Cy3 or Alexa647 labeled target oligonucleotide yielded FRET-sensitized emission from the dyes, which was used as the analytical signal. The green QD–Cy3 FRET
pair utilized the SMN1 sequence and the red QD–Alexa647 pair utilized the LacZ sequence. (b) A cartoon of the expected emission profiles, where the bracketed
regions are of particular analytical interest. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)

(spots) [19–25]. Techniques based on encoded microspheres,
including those encoded with QDs, have recently allowed multi-
plexed analyses of nucleic acids using a single excitation source,
but require observation of individual microspheres, typically by
flow cytometry [26–30]. By combining the broad absorption
spectra of quantum dots with stimulated emission via FRET,
the new work reported herein has developed an approach to
multi-colour nucleic acid diagnostics which uses only a sin-
gle excitation source, permits bulk measurements, and does
not require discrete sensing elements for each target. Multi-
plexed schemes combining FRET with quantum dots have been
characterized [31] and applied [7] to non-nucleic acid targets
previously. However, in these cases the acceptor species was
a dark quencher rather than a fluorophore. Although effective,
the approach using a quencher may be susceptible to non-FRET
mechanisms of luminescence quenching, potentially resulting
in spurious signals. The proposed emission scheme is advanta-
geous in that only close proximity between the acceptor and the
donor QD can result in emission via FRET.
Using a single excitation wavelength near the ultraviolet
region of the spectrum, different sizes of QDs can be simultane-
ously excited due to their broad absorption spectra. In this work,
smaller green emitting QDs are conjugated to probe oligonu-
cleotides that are complementary to target sequences diagnos-
tic of a genetic disorder (spinal muscular atrophy). Similarly,
larger red emitting QDs are conjugated to probes diagnostic
of a pathogen (E. coli). Upon hybridization of the green and
red QD–DNA conjugates with target sequences labeled with
Cyanine 3 (Cy3) and Alexa Fluor 647 (Alexa647), respectively,
the QDs act as energy donors. As shown in Fig. 1, the result
is simultaneous FRET-sensitized emission from the Cy3 and
Alexa647 acceptor dyes. Aside from the ability to perform
multi-colour analyses on a single sensor element with a single
excitation source, the resistance of this scheme to photobleach-
ing is similar to that of QDs since the organic fluorophores are
excited indirectly via FRET. This work provides demonstration
of proof-of-principle for both single-colour and two-colour anal-
yses in solution phase experiments using a sensitized emission
scheme. The quantum yields and photoluminescence lifetimes
of QD–DNA conjugates in this system are characterized. Non-
specific adsorption of oligonucleotides on the QD surface lead-
ing to FRET is encountered, however, the use of an intercalating
dye such as ethidium bromide is shown to alleviate this issue.
Aside from their inherent selectivity for double-stranded DNA,
intercalating dyes may also be a better pragmatic approach since
they avoid labeling of target material.
2. Experimental
2.1. Reagents and oligonucleotides
Adirondrack Green and Maple Red CdSe/ZnS core/shell
semiconductor nanocrystals (QDs) in toluene were from Evi-
dent Technologies (Troy, NY, USA). Ninety-eight percent mer-
captoacetic acid, 99.5% N,N-diisopropylethylamine, and N-
(3-dimethylaminopropyl)-N -ethylcarbodiimide hydrochloride
 W.R. Algar, U.J. Krull / Analytica Chimica Acta 581 (2007) 193–201
Table 1
Oligonucleotide sequences used in hybridization assays
Oligonucleotide Sequence
Green QD-SMN1
Probe NH2 C6 H12 -5 -ATT TTG TCT GAA ACC CTG
T-3
Target-Cy3 Cy3-5 -ACA GGG TTT CAG ACA AAA T-3
Target 5 -ACA GGG TTT CAG ACA AAA T-3
Red QD-LacZ
Probe NH2 C6 H12 -5 -CTT ACT TCC ATG ATT TCT
TTA ACT-3
Target-Alexa647 Alexa647-5 -AGT TAA AGA AAT CAT GGA
AGT AAG-3
(EDC) were from Sigma–Aldrich (Oakville, Ont., Canada) and
used without further purification. Chloroform was from EM Sci-
ence (Toronto, Ont., Canada) and used as received.
Modified and unmodified oligonucleotides were from Inte-
grated DNA Technologies (Coralville, IA, USA) and were
HPLC purified by the manufacturer. The oligonucleotides were
dissolved in deionized water with a specific resistance of
18 M cm−1 . Nucleotide sequences are listed in Table 1. Ethid-
ium bromide homodimer was obtained from Sigma–Aldrich.
Water was deionized and purified by the Milli-Q cartridge purifi-
cation system (Millipore Corp., Mississauga, Ont., Canada).
Tris–borate (TB, 90 mM, pH 7.4) buffer solutions were prepared
with autoclaved double-distilled water and filtered through a
0.2 ␮m syringe filter for the preparation of QD solutions.
2.2. Instruments
Ultraviolet–visible absorption spectra were measured using
a Libra S22 spectrometer (Bichrom Ltd., Cambridge, UK) and a
HP 8452A Diode-Array Spectrometer (Hewlett Packard Cor-
poration, Palo Alto, CA, USA). Solution phase fluorescence
spectra were measured using a QuantaMaster PTI Spectroflu-
orimeter and Felix Software (Photon Technology International,
Lawrenceville, NJ, USA). Photoluminescence lifetime measure-
ments were made using a time correlated single photon counter
(constructed in-house), driven by a 520 nm femtosecond laser
(pulse duration: 200 fs, repetition rate: 15 MHz, bandwidth:
3 nm, mean power: 1 mW at 520 nm).
2.3. Quantum dot surface modification and preparation of
DNA conjugates
QDs in toluene were made water soluble by ligand exchange
with mercaptoacetic acid (MAA). In a typical procedure, QDs
in toluene were diluted in chloroform with >5 × 104 -fold excess
of MAA and sufficient N,N-diisopropylethylamine to render
the solution basic. The mixture was sonicated for 2–3 min and
refluxed under argon for 8–12 h. During this period the quan-
tum dots precipitated. As has been reported elsewhere [32],
the heating was found to be essential for producing dots that
could withstand purification. The precipitate and supernatant
were centrifuged to produce a compact pellet, the supernatant
discarded, and the precipitate was washed three times with
195
chloroform. Each wash consisted of mixing, centrifuging, and
discarding the supernatant. After drying in air to remove resid-
ual chloroform, the precipitate was dissolved in TB buffer. This
was followed by re-precipitation via the addition of 95% ethanol
(ca. 3:1 EtOH:TB) and centrifugation. This step was repeated
once more prior to dissolving the water soluble quantum dots
(MAA–QDs) in the desired amount of TB buffer. The concentra-
tion of MAA–QDs was determined by absorption spectroscopy
using the first absorption peak (green = 515 nm; red = 600 nm).
Typical recoveries from ligand exchange were 80–90%, with the
larger red QDs having higher recoveries.
DNA conjugates were prepared by mixing MAA–QDs with n
= 1 or 2 equiv. of amine modified oligonucleotides in TB buffer
containing EDC. Reaction mixtures were generally 5–20 ␮M
in QDs and allowed to stand 6–8 h at room temperature. It
should be noted that phosphate buffers commonly used with
oligonucleotides interfered with the coupling reaction. Follow-
ing the reaction, excess EDC was removed by precipitation of the
QD–DNA conjugate with ethanol, centrifugation, and discard-
ing the supernatant. The conjugates were then redissolved in TB
buffer and precipitated with ethanol twice more before finally
dissolving in the desired amount of buffer. The probe-to-QD
ratio was controlled by the stoichiometry of the EDC cou-
pling reaction. The properties of the resulting conjugates were
assumed to be dominated by a large population of QD–n × DNA
conjugates with minimal influence from the potential existence
of small QD–(n ± 1) × DNA conjugate populations. Typical
conjugate recoveries were 70–80% and final concentrations
were determined by UV–visible absorption spectroscopy. As
reported elsewhere [33], we found that the QDs were suscep-
tible to aggregation during the EDC coupling reaction. This
appeared to be a function of the amount of EDC, and was ame-
liorated by reducing the quantity of EDC in solution, albeit still
working with a 104 to 105 -fold excess. Green QD–DNA conju-
gates were centrifuged for 15 min at 10 000 rpm; red QD–DNA
conjugates were centrifuged for 6–8 min at 6000 rpm. The latter
was required due to the greater instability of the red QDs, which
would not consistently re-disperse if centrifuged too vigorously.
2.4. Determining quantum yields, lifetimes, and Förster
distances
The Förster distance, Eq. (1), is a characteristic of a
donor–acceptor pair, and depends on factors including the
refractive index of the surrounding medium, n, the donor quan-
tum yield, ΦD , the relative orientation between donor emission
and acceptor absorption dipoles, and the degree of spectral reso-
nance between the two species [34]. These latter two parameters
are described by the orientation factor, κ2 , and spectral overlap
integral, J, respectively. The spectral overlap integral, Eq. (2),
is a function of the fluorescence intensity of the donor, FD , and
molar absorptivity of acceptor, εA , as a function of wavelength,
λ, normalized against the total donor emission [34]. Acceptor
ultraviolet–visible absorption and donor fluorescence emission
spectra were obtained with 3 ␮M solutions of oligonucleotide,
and 10 ␮M (green) and 1.0 ␮M (red) solutions of QD, respec-
tively. From the spectra obtained, the integrands in Eq. (2) were
196 W.R. Algar, U.J. Krull / Analytica Chimica Acta 581 (2007) 193–201
calculated at 0.5 nm increments and integrated numerically to
determine the spectral overlap.
R6o = 8.79 × 10−28 mol × (n−4 κ2 ΦD J) (1)
 were acquired using 396 nm excitation and TB buffer served
FD (λ)εA (λ)λ4 dλ
J=  (2)
FD (λ) dλ
Quantum yield values for QDs were determined relative to
fluorescein dye in sodium borate buffer fixed at pH 9.5. The
quantum yield of fluorescein under these conditions is known to
be 0.93 [35] and the quantum yield, Φ, of QDs were determined
as a ratio, Eq. (3a), of their integrated emission, F dλ, corrected
for different molar absorptivities at the wavelength of excitation,
ε, and for concentration, c. Fluorescein was excited at 490 nm. To
minimize error, fluorescence and absorbance, A, were measured
from the same aliquot of solution, in which case Eq. (3a) reduces
to Eq. (3b) via Beer’s law. A consistent path length is assumed
in both equations.
 dA20 oligonucleotides or salmon sperm DNA were conducted
F dλ ε Φ c
 = (3a)
Fref dλ εref Φref cref
 ium bromide was detectable. The excitation wavelength was
F dλ Aref
Φ = Φref  (3b)
A Fref dλ
Photoluminescence lifetimes were measured using 1.0 or
0.06 ␮M solutions of the desired green or red QD/QD–DNA con-
jugate, respectively. Decay curves were obtained using SPCM
software (Version 8.50) and SPC-630 hardware (Becker & Hickl
GmbH, Berlin, Germany), and analyzed using SPCImage (Ver-
sion 2.8.3.2921, Becker & Hickl GmbH). Decay curves were fit
with a monoexponential lifetime by minimizing the χ2 value.
The system response used in the fitting routine was measured
experimentally. A monoexponential decay fit the data well and
simplified analysis of the FRET efficiency. The emission of the
green QD was isolated by using a combination of a 530 nm
long-pass coloured glass filter (Melles Griot, Rochester, NY,
USA) and 550 nm short-pass interference filter (ThorLabs, New-
ton, NJ, USA). Using a spectrofluorimeter, this was confirmed
to block the majority of the Cy3 or ethidium bromide emis-
sion excited directly at 520 nm. The red QD emission was iso-
lated using a combination of a 590 nm long-pass filter (Nikon,
Kawaski, Japan) and a 650 nm short-pass interference filter
(ThorLabs).
2.5. Hybridization assays
For single-colour experiments, green QD–1 × DNA and
red QD–2 × DNA conjugates were prepared using the probe
oligonucleotides listed in Table 1. Solutions were prepared as
1.0 and 0.06 ␮M in green and red QD-conjugates, respectively,
in TB buffer with the desired equivalents of labeled target. Solu-
tions were allowed to stand at room temperature for 7–8 h prior
to measurement. Measurements on the red system were obtained
with a 615 nm long-pass coloured glass filter (Melles-Griot). The
green system required no optical filtering.
For multi-colour experiments, green QD–1 × DNA and red
QD–2 × DNA conjugates were prepared as a mixture in TB
buffer with 1.0 and 0.06 ␮M concentrations, respectively. Tar-
gets were introduced at the desired concentrations and the solu-
tions were allowed to stand for 7–8 h prior to measurement.
Both single-colour and two-colour luminescence spectra
as the blank. Direct excitation of Cy3 is minimized at 385 nm
and direct excitation of Alexa647 is minimized at 425 nm. The
minimum additive emission from direct excitation with a mix-
ture of the two dyes was found at 396 nm. In these experiments,
direct excitation of either dye at 396 nm at any of the concentra-
tions used was not distinguishable from the background within
the experimental precision.
Single-colour experiments with ethidium bromide were con-
ducted similarly to those described above, except that the
target oligonucleotides were unlabeled and 6 equiv. of ethid-
ium bromide were added after the hybridization period. The
ethidium bromide was allowed to equilibrate 3–4 h prior to
measurement. Experiments involving mixtures of target and
similarly, except for the additional nucleic acid material. In
contrast to Cy3 and Alexa647, the direct excitation of ethid-
selected to be 400 nm were the lowest direct excitation of
ethidium bromide was observed compared to other excitation
wavelengths.
3. Results and discussion
3.1. Characterization of quantum dots and QD–DNA
conjugates
Changes in QD quantum yield associated with ligand
exchange and DNA conjugation are given in Table 2. The quan-
tum yield of the QDs decreased by roughly an order of magnitude
between the organic and aqueous systems, and further changed
upon DNA conjugation. A 2 nm shift to longer wavelengths was
observed in the emission of the green QDs between organic and
aqueous phases. While not listed in Table 2, it was also found that
increasing the number of probes per green QD from one to two
approximately doubled the quantum yield. Similarly, an approx-
imate two-fold increase in quantum yield was also observed
when there were six probes per red QD rather than two. These
results suggest that the conjugation of DNA helps passivate the
QD surface and improve quantum yield, but that the effect may
be partially counteracted by the additional purification process
associated with removal of excess EDC. The quantum yield of
the water soluble QDs and QD–DNA conjugates appeared to
vary significantly between preparations. The uncertainties listed
in Table 2 represent the variability between different batch prepa-
rations. Replicate measurements of samples from a single batch
show variation in the range of 0.1–4%. Given the well-known
instability of thiol capped QDs in aqueous solution, it is not
surprising that the quantum yield was sensitive to the nature of
the preparation. It is suspected that the purification steps are
the main source of the variability since the equilibrium between
bound and unbound thiol is disrupted and not re-established until
the final dispersion in Tris–borate (TB) buffer.
 W.R. Algar, U.J. Krull / Analytica Chimica Acta 581 (2007) 193–201
Table 2
Quantum yields and Förster distances of QDs
QD colour Quantum yield (×10−2 )
QDs (toluene) MAA–QDs (aq.)b
Greena 28.6 ± 0.8 (524 nm) 1.8 ± 0.6 (526 nm)
Reda 9.2 ± 0.4 (606 nm) 0.4 ± 0.1 (606 nm)
a The position of the emission maximum, λmax , is given in parentheses.
b Water soluble QDs with mercaptoacetic acid ligands.
c The green QD have a single conjugated oligonucleotide; the red QDs have two conjugated oligonucleotides.
3.2. Single-colour hybridization assays with Cy3 and
Alexa647
A hybridization event between a QD–DNA conjugate and a
dye-labeled complementary sequence of DNA brought the dye
label within a distance on the order of the Förster radius of the
QD–dye FRET pair. As a consequence, FRET-sensitized dye
fluorescence was observed. In these experiments, green and red
emitting CdSe/ZnS core-shell QDs were used as energy donors
with Cy3 and Alexa647 labeled oligonucleotides as energy
acceptors. The absorption and emission spectra for these QDs
and dyes are shown in Fig. 2.
Hybridization between probe and target has been confirmed
experimentally in a number of ways. Significant differences
in the kinetic rates of hybridization and non-specific adsorp-
tion using a non-complementary sequence were observed,
and fluorescence spectra obtained using PicoGreen to stain
double stranded DNA indicated the formation of hybrids
(see Supplementary Information). Similarly, ethidium bromide
demonstrated the expected fluorescence lifetime of ca. 20 ns
(characteristic of the presence of dsDNA) when QD–DNA con-
Fig. 2. Spectral overlap (shaded) for the two FRET systems of interest: (a)
a green QD donor with a Cy3 labeled acceptor and (b) a red QD with an
Alexa647 labeled acceptor. Both the normalized absorption (abs.) and emis-
sion (em.) spectra are shown. Note that the absorption coefficient for the red QD
is approximately an order of magnitude larger than for the green QD. (For inter-
pretation of the references to colour in this figure legend, the reader is referred
to the web version of the article.)
197
Förster distance (nm)
QD–DNA conjugate (aq.)c Cy3 Alexa647
0.4 ± 0.2 (526 nm) 2.7 ± 0.2 –
0.9 ± 0.6 (606 nm) – 3.52 ± 0.2
jugates were mixed with complementary material. Melt curves
were also obtained with QD–DNA conjugates demonstrating
dehybridization as a function of temperature.
The hybridization-FRET-sensitized fluorescence of both Cy3
and Alexa647 labeled targets is shown in Fig. 3 as a func-
tion target concentration. In Fig. 3a, solutions containing green
QD with one conjugated SMN1 probe oligonucleotide (green
QD–1 × DNA) were exposed to different amounts of SMN1
target. In Fig. 3b, solutions containing red QD with two con-
jugated LacZ oligonucleotide probes (red QD–2 × DNA) were
exposed to different amounts of LacZ target. In both cases, the
acceptor fluorescence was observed to increase with increas-
ing target hybridization. The fluorescence intensity in the Cy3
spectral region was linear at target concentrations ≥400 nM and
varied predictably at lower concentrations. For a 1.0 ␮M con-
centration of QD–DNA conjugate, the limit-of-detection (LOD)
is estimated to be 40 nM using the definition of three standard
deviations above the baseline in the absence of target. The fluo-
rescence intensity in the Alexa647 spectral region was found
to change linearly at all target concentrations, with an esti-
mated LOD of 12 nM for a QD–DNA conjugate concentration of
0.06 ␮M. For both colours, the upper limit of the dynamic range
was defined by the concentration of the QD–DNA conjugate.
The different LODs between the green and red systems demon-
Fig. 3. Single-colour experiments demonstrating the ability to quantitatively
detect labeled target oligonucleotide sequences via FRET-sensitized emission:
(a) a 1.0 ␮M solution of green QD–1 × DNA conjugate with 80, 100, 200, 400,
600, 800, and 1000 nM of Cy3-labeled target; (b) a 0.06 ␮M solution of red
QD–2 × DNA conjugate with 20, 40, 60, 80, 100, and 120 nM of Alexa647-
labeled target.
198 W.R. Algar, U.J. Krull / Analytica Chimica Acta 581 (2007) 193–201
strate that the lower limit of the dynamic range, and thus LOD,
is a function of the amount of QD–DNA conjugate. Although
the dynamic range is small at one particular concentration of
conjugate, it is effectively tunable over a range of QD–DNA
conjugate concentrations. For example, a reduction in green QD
concentration of two orders of magnitude is still detectable in
a simple spectrofluorimeter and provides picomolar LODs. In
addition, FRET between quantum dots and organic dyes has
been shown to be detectable at the single molecule level [36],
suggesting that this diagnostic methodology could be extended
to such a platform. In terms of ensemble measurements, any lack
of sensitivity compared to other techniques may be offset by this
resistance of this scheme to photobleaching, making it ideal for
a continuous monitoring biosensor application.
It should be noted that the difference in concentration
between the green and red systems is required to give similar
QD luminescence intensities. In addition, the quantum yield of
the Alexa647 dye was less than that of Cy3 in this experiment.
As a consequence of these two factors, the large photolumines-
cence from the red QD largely obscured the FRET-sensitized
shoulder from Alexa647 emission. However, it was possible to
reduce the emission contribution of the red QD with the addition
of a 615 nm long-pass glass filter. This had the effect of shift-
ing the apparent emission maximum from 606 to 620 nm (see
Supplementary Material for spectra). All FRET-related steady
state data were obtained in this manner and the red shift of the
QD emission maximum in Fig. 3 relative the unfiltered emission
in Fig. 2 is a consequence of this effect.
Photoluminescence lifetimes measured for the QD–DNA
conjugates are listed in Table 3 and may be compared to the life-
times measured for green (7.0 ± 0.2 ns) and red (8.4 ± 0.2 ns)
MAA–QDs. Changes in lifetime and relative quantum yield
associated with the hybridization of an unlabeled complemen-
tary target oligonucleotide (QD–n × dsDNA) were observed.
The increase in quantum yield upon hybridization is consistent
with the aforementioned increase in quantum yield with a greater
number of conjugated probe oligonucleotides. The standard
deviations reported in Table 3 are associated with measurements
using a single batch preparation of QD–DNA conjugates. Since
the quantum yield and lifetime may change with the nuances of
each conjugate preparation, it is not the absolute values which
are of interest, but rather the relative change in those values. It is
important to note that the changes in quantum yield and lifetime
observed in Table 3 are associated entirely with the hybridiza-
Table 3
FRET efficiency derived from changes in quantum yield and photoluminescence lifetime
Green systema Lifetime (ns) Relative QYb
QD–1 × DNA 8.5 ± 0.3 1.00 ± 0.05
QD–1 × dsDNA 8.7 ± 0.3 1.31 ± 0.01
QD–1 × dsDNA–Cy3 4.4 ± 0.1 0.48 ± 0.03
FRET efficiency (%)c 52 ± 3 63 ± 3
a QD–n × DNA represents QDs with n conjugated oligonucleotide probes, which may by hybridized with target as indicated by “dsDNA”. The target may also be
fluorophore labeled, as indicated.
b Relative quantum yield.
c FRET efficiency as a percentage.
tion process since there is no clean-up step (as done with probe
conjugation with EDC).
The FRET efficiencies observed for the green and red sys-
tems are 52 and 6.7%. These values were calculated from the
data in Table 3 by comparing the lifetimes (and relative quan-
tum yields) for QD–n × dsDNA and QD–n × dsDNA–acceptor
dye. The relationships between FRET efficiency, E, and life-
time, τ, or quantum yield, Φ, are given by Eq. (4), where DA
indicates a donor quantity (D) in the presence of n acceptors
(A). The efficiencies calculated from the lifetime/quantum yield
data are listed in Table 3, and allow the donor–acceptor sepa-
ration, r, to be determined from Eq. (5). The Förster distances,
Ro , are listed in Table 2. The Ro values are calculated from
spectra and the quantum yields for QD–DNA conjugate, based
on a buffer refractive index of 1.34 and an assumed orientation
factor of κ2 = 2/3. This last approximation has been shown to
be valid when the orientation of both the quantum dot exci-
ton and the acceptor transition dipole are expected to be at
least partially randomized [37]. The spectral overlaps deter-
mined for the green QD–Cy3 and red QD–Alexa647 pairs are
(5.0 ± 0.4) × 10−10 cm6 and (1.14 ± 0.07) × 10−9 cm6 , respec-
tively. The donor–acceptor distances calculated from Eqs. (4)
and (5), and the lifetime (or quantum yield) data are 2.7 nm (or
2.5 nm) and 6.1 nm (or 5.6 nm), respectively. These values are
quite reasonable considering the hydrodynamic radius of DNA,
the QD dimensions (green QD: 2.1 nm core diameter; red QD:
5.2 nm core diameter [38]), and assuming the DNA adopts some
conformation along the surface of the QD. Given the large size
of the red QD, it is not surprising that the FRET efficiency is
much higher in the green system.
τDA ΦDA
E =1− =1− (4)
τD ΦD
nR6o
E= (5)
nR6o + r 6
3.3. Simultaneous two-colour hybridization assay with Cy3
and Alexa647
The single-colour green and red QD systems were success-
fully integrated into a two-colour detection scheme as shown in
Fig. 4. The FRET-sensitized signals in the Cy3 and Alexa647
emission windows were observed to change as a function of the
quantity of target material. Fig. 4(a) shows the change in FRET
Red systema Lifetime (ns) Relative QYb
QD–2 × DNA 6.2 ± 0.2 1.00 ± 0.03
QD–2 × dsDNA 5.9 ± 0.1 1.34 ± 0.03
QD–2 × dsDNA–Alexa647 5.5 ± 0.1 1.18 ± 0.18
FRET efficiency (%)c 6.7 ± 0.2 11 ± 2
 W.R. Algar, U.J. Krull / Analytica Chimica Acta 581 (2007) 193–201
Fig. 4. Two-colour experiments demonstrating the ability to quantitatively
detect two different labeled oligonucleotide sequences simultaneously via
FRET-sensitized emission: (a) 1.0 and 0.06 ␮M solution of gQD–1 × DNA
and rQD–2 × DNA, respectively, with: no targets, 330 nM Cy3-target + 120 nM
Alexa647-target, 660 nM Cy3-target + 80 nM Alexa647 target, and 1000 nM
Cy3-target + 40 nM Alexa647-target; (b) 1.0 and 0.06 ␮M solution of
gQD–1 × DNA and rQD–2 × DNA, respectively, with: no targets, 330 nM Cy3-
target + 40 nM Alexa647-target, 660 nM Cy3-target + 80 nM Alexa647 target,
and 1000 nM Cy3-target + 120 nM Alexa647-target. The green and red systems
appeared to change independently. Channels are delineated by the dashed line
at 600 nm.
signals when the green target concentration is increased and
red target concentration is simultaneously decreased. Fig. 4(b)
shows the change in FRET signals when both the green and
red target concentrations are simultaneously increased. Com-
paring Fig. 4(a) and (b), it is observed that the FRET-sensitized
signals changed independently of one another. The data clearly
demonstrates the capacity for a two-colour detection scheme.
The data in Fig. 4 also shows that 1.0 and 0.06 ␮M concen-
trations of green and red QD–DNA conjugate were able to
detect target concentrations in the range of 10−1 to 100 ␮M
and 101 to 102 nM, respectively. However, it should again be
noted that the dynamic range of either single-colour system
can be tuned by changing the quantity of QD-conjugate and
is limited only by the ability to resolve the four emission com-
ponents and the effects of non-specific adsorption (discussed in
Section 3.4).
3.4. Non-specific adsorption
Non-specific adsorption of oligonucleotides has been found
to be very strong on MAA–QDs and only slightly reduced
with QD–DNA conjugates. The FRET signal may be semi-
quantitatively treated as the ratio of the emission intensity at
the Cy3 emission maximum to the intensity at the green QD
maximum. Ratios of roughly 0.1, 1.8, and 1.0 were obtained for
green QD–1 × DNA, green QD–1 × DNA hybridized with fully
complementary target, and green QD–1 × DNA with adsorbed
non-complementary target, respectively, indicating a signal-to-
noise ratio less than two (S/N < 2). The red QD–2 × DNA system
showed slightly stronger non-specific adsorption, likely due to
199
the larger surface area available. In an effort to prevent non-
specific adsorption of oligonucelotides, maintain quantum yield,
and allow facile hybridization, the use of bovine serum albumin
(BSA), polyvinyl pyrrolidone (PVP), poly-l-lysine as blocking
agents, or additions of Tween 20 or a small percentage of for-
mamide were explored. Only BSA was successful in preventing
non-specific adsorption, but the hybridization signal from com-
plementary target was found be greatly reduced.
An interesting aspect seen in Fig. 4 is that the Cy3 and
Alexa647 FRET signals in the two-colour experiments were less
than those observed in the single-colour experiments. Although
the hybridization signals changed independently with target con-
centration, this result may indicate that there was interaction
between the two systems. Non-specific adsorption in the two-
colour system was observed to be less that that seen in the single-
colour experiments. Although the green QDs were approxi-
mately ≥17-fold more concentrated than the red quantum dots,
the ratio of red-to-green QD surface area is >6 (based on core
diameter) and could have helped offset the concentration differ-
ence. Adsorption of red target on green quantum dots and vice
versa could not be directly detected due to the absence of signif-
icant spectral overlap, however, the effect may be observable as
the reduction of the red signal associated with hybridization. As
can be seen by comparing Fig. 4 with Fig. 3, adsorption of green
target onto red QDs appears to have less effect on the spectra.
The different extents of the cross-adsorption effects on the green
(smaller effect) and red (larger effect) channels were likely due,
in part, to the concentration difference in target oligonucleotides.
As shown in Fig. 5, preliminary experiments also demonstrate
that the hybridization kinetics of the red QD–probe conjugates
were substantially faster than the green QD–probe conjugates,
allowing red probe–target hybridization on the red QDs to par-
tially offset the adsorption of red target on the green QDs. As
Fig. 5. Normalized kinetic traces for the evolution of FRET-sensitized Cy3
fluorescence (at 560 nm) from (a) 1.0 ␮M complementary target with 1.0 ␮M
of green QD–1 × DNA conjugate and (b) 0.12 ␮M complementary target with
0.06 ␮M of red QD–2 × DNA conjugate. Note the much faster kinetics of (b)
relative (a). (For interpretation of the references to colour in this figure legend,
the reader is referred to the web version of the article.).
200 W.R. Algar, U.J. Krull / Analytica Chimica Acta 581 (2007) 193–201
Fig. 6. Experiments demonstrating the ability to detect labeled target oligonu-
cleotide sequences via FRET-sensitized ethidium bromide emission: (a) a
1.0 ␮M solution of green QD–1 × DNA conjugate with 6.0 ␮M ethidium bro-
mide and (i) 0.0 ␮M target, (ii) 1.0 ␮M non-complementary sequence, (iii)
0.2 ␮M target, (iv) 0.4 ␮M target, (v) 0.6 ␮M target, (vi) 1.0 ␮M target; (b)
a 1.0 ␮M solution of green QD–2 × DNA conjugate with 12.0 ␮M ethidium
bromide and (i) 0.0 ␮M target, (ii) 2.0 ␮M non-complementary sequence, (iii)
4.0 ␮M non-complementary, (iv) 0.8 ␮M target, (v) 1.2 ␮M target, (vi) 1.6 ␮M
target, (vii) 2.0 ␮M target. In (a), the curve for 2.0 ␮M of non-complementary
target has been omitted for clarity, but is only 3% greater than that for 0.2 ␮M
target. The curve for 0.4 ␮M target in (b) has also been omitted for clarity. Curves
for non-complementary material are shown as dashed lines. The data suggests
the signal associated with hybridization is 9–10-fold larger than that of an equal
amount of non-complementary material. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of the
article.)
a consequence, the red FRET signals in the two-colour system
are not completely diminished.
3.5. Hybridization assay with ethidium bromide
Ethidium bromide is an intercalating dye which shows a
strong enhancement of quantum yield upon incorporation into
double-stranded DNA, but shows much weaker fluorescence in
the presence of single-stranded DNA. Fig. 6 shows hybridiza-
tion assays with green QD systems: (a) 1.0 ␮M of green
QD–1 × DNA with the addition of variable amounts of non-
labeled target or non-complementary sequence and 6 equiv. of
ethidium bromide; (b) 1.0 ␮M of green QD–2 × DNA with vari-
able amounts of target or non-complementary sequence and
12 equiv. of ethidium bromide (i.e. 6 equiv. per probe). The
figures also show the direct excitation of ethidium bromide asso-
ciated with the system in the absence of target. Six equivalents
of ethidium bromide allows the maximum acceptor absorp-
tion cross-section for FRET (as per Eq. (5)) while ensuring
that the intercalative capacity of the 19-base SMN1 sequence
is not exceeded. The use of multiple equivalents also helps
offset the lower quantum yield and molar absorption coeffi-
cient of ethidium bromide as compared to Cy3. As observed in
Fig. 6, increasing FRET-sensitized ethidium bromide emission
with increasing target is observed and signal-to-noise is greatly
increased with respect to non-specific adsorption, reaching a
value of S/N = 9–10. This is seen in Fig. 6a, where the signal
from 1.0 equiv. of non-complementary material is less than that
observed with 0.2 equiv. of target. In addition, the signal asso-
ciated with 2.0 equiv. of non-complementary material is equal
(within experimental precision) to that observed with 0.2 equiv.
of target, suggesting the upper limit of S/N. Similar S/N is
observed with the green QD–2 × DNA system shown in Fig. 6b.
In this case, the signal from 4.0 equiv. of non-complementary
material equal to the signal expected for 0.4 equiv. of target. The
estimated LOD for the green QD–1 × DNA–ethidium bromide
system is 80 nM at a QD-conjugate concentration of 1.0 ␮M.
The estimated LOD for the green QD–2 × DNA–ethidium bro-
mide system is 165 nM. The upper limits of the dynamic ranges
are 1.0 and 2.0 ␮M, respectively. Due to the lower quantum
yield of ethidium bromide relative Cy3, the tunability of the
dynamic range is roughly an order of magnitude less than that
associated with Cy3. The system has been tested against matri-
ces containing a six-fold excess of non-complementary dA20
oligonucleotide and 10-fold excess of salmon sperm DNA. The
hybridization signals were approximately 80 and 100% of that
associated with clean matrices, respectively.
Lifetime measurements suggest that the FRET efficiencies
for the QD–ethidium bromide systems are approximately 5%.
The lower efficiency compared with Cy3 is due to the much
lower molar absorption coefficient of ethidium bromide. The
calculated Förster distance for the green QD–ethidium bromide
pair is 1.82 ± 0.08 nm. With judicious choice of a second inter-
calating dye, it should be possible to construct a two-colour
intercalating dye system similar to that demonstrated with Cy3
and Alexa647, but less sensitive to non-specific adsorption.
Dyes potentially suitable for use with the red QD include TO-
PRO-3 and YO-PRO-3, which both have spectrally resolved
emission from the red QD and absorption shoulders in over-
lap with the red QD emission. However, since these dyes have
no sequence selectivity, each probe/target hybrid would have
mixture of acceptors. This would not result in spurious FRET
signals since only one dye will have significant spectral overlap
with the donor, but a reduction in overall signal intensity would
occur. The use of probe-tethered intercalating dyes [39–41]
would address this issue. As discussed previously with respect
to the Cy3 and Alexa647 labeled targets, moving the scheme to
a single-molecule platform to improve sensitivity should be fea-
sible. Similarly, in a continuous monitoring biosensor scheme,
the QD–ethidium bromide FRET system should remain substan-
tially more resistant to photobleaching than in a direct excitation
scheme.
4. Conclusions
The potential for both single- and two-colour diagnostic
schemes for nucleic acid hybridization based on FRET between
quantum dot donors and acceptor fluorophore labeled oligonu-
cleotides has been demonstrated. The single-colour schemes
were quantitative, where FRET-sensitized acceptor fluorescence
was observed to change systematically as a function of target
concentration. The dynamic range is tunable with the concen-
tration of QD–DNA conjugate (within the limits of available
 W.R. Algar, U.J. Krull / Analytica Chimica Acta 581 (2007) 193–201
instrument). Results show that the green QD–Cy3 pair can
detect target as low as 4% as the QD concentration. The red
QD–Alexa647 pair can detect target as low as 10% of the QD
concentration. The green QD–ethidium bromide pair can detect
target as low as 8% of the QD concentration. In all cases, the
upper limit of the dynamic range is defined by the QD con-
centration. In the two-colour scheme, the colours responded
proportionately to target concentration and relatively indepen-
dently of one another, although with reduced signals relative the
single-colour system. This was a consequence of non-specific
adsorption, which was not preventable with standard blocking
agents. To alleviate the effects of non-specific adsorption, the
use of an intercalating dye (ethidium bromide) as an acceptor
was demonstrated. Signal-to-noise with respect to hybridization
and non-specific adsorption increased from S/N < 2 with the Cy3
or Alexa647 systems to S/N = 9–10 with ethidium bromide. The
ethidium bromide system was also found to effectively detect
target in a matrix containing an excess of non-complementary
oligonucleotide or salmon sperm DNA.
Acknowledgements
The authors acknowledge the Natural Sciences and Engineer-
ing Research Council of Canada (NSERC) for financial support
of this work and provision of a fellowship to WRA. The authors
also thank Prof. Virginijus Barzda, Dr. Arkady Major, and Dr.
Paul Piunno for the use of their instrumentation and assistance
in acquiring the lifetime data.
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at doi:10.1016/j.aca.2006.08.026.
References
[1] M. Han, X. Gao, J.Z. Su, S. Nie, Nat. Biotechnol. 19 (2001) 631.
[2] J.K. Jaiswal, H. Mattoussi, J.M. Mauro, S.M. Simon, Nat. Biotechnol. 21
(2003) 47.
[3] M. Stroh, J.P. Zimmer, D.G. Duda, T.S. Levchenko, K.S. Cohen, E.B.
Brown, D.T. Scadden, V.P. Torchilin, M.G. Bawendi, D. Fukumura, R.K.
Jain, Nat. Med. 11 (2005) 678.
[4] A.M. Smith, X. Gao, S. Nie, Photochem. Photobiol. 80 (2004) 377.
[5] I.L. Medintz, H.T. Uyeda, E.R. Goldman, H. Mattoussi, Nat. Mater. 4
(2005) 435.
[6] F. Pinaud, X. Michalet, L.A. Bentolila, J.M. Tsay, S. Doose, J.J. Li, G. Iyer,
S. Weiss, Biomaterials 27 (2006) 1679.
[7] E.R. Goldman, A.R. Clapp, G.P. Anderson, H.T. Uyeda, J.M. Mauro, I.L.
Medintz, H. Mattoussi, Anal. Chem. 76 (2004) 684.
201
[8] E.R. Goldman, I.L. Medintz, J.L. Whitley, A. Hayhurst, A.R. Clapp, H.T.
Uyeda, J.R. Deschamps, M.E. Lassman, H. Mattoussi, J. Am. Chem. Soc.
127 (2005) 6744.
[9] J.H. Kim, D. Morikis, M. Ozkan, Sens. Actuators B 102 (2004) 314.
[10] C.Y. Zhang, H.C. Yeh, M.T. Kuroki, T.H. Wang, Nat. Mater. 4 (2005) 826.
[11] R. Gill, I. Willner, I. Shweky, U. Banin, J. Phys. Chem. B 109 (2005) 23715.
[12] D. Zhou, J.D. Piper, C. Abell, D. Klenerman, D.J. Kang, L. Ying, Chem.
Comm. (2005) 4807.
[13] E. Chang, J.S. Miller, J. Sun, W.W. Yu, V.L. Colvin, R. Drezek, J.L. West,
Biochem. Biophys. Res. Commun. 334 (2005) 1317.
[14] I.L. Medintz, A.R. Clapp, H. Mattoussi, E.R. Goldman, G. Fisher, J.M.
Mauro, Nat. Mater. 2 (2003) 630.
[15] A. Almadidy, J. Watterson, P.A.E. Piunno, S. Raha, I.V. Foulds, P.A. Hor-
gen, A. Castle, U. Krull, Anal. Chim. Acta 461 (2002) 37.
[16] S. Ahn, D.R. Walt, Anal. Chem. 77 (2005) 5041.
[17] J.H. Watterson, S. Raha, C.C. Kotoris, C.C. Wust, F. Gharabaghi, S.C.
Jantzi, N.K. Haynes, N.H. Gendron, U.J. Krull, A.E. Mackenzie, P.A.E.
Piunno, Nucl. Acids Res. 32 (2004) e18.
[18] I. Biran, D.R. Walt, Anal. Chem. 74 (2002) 3046.
[19] M. Schena, D. Shalon, R.W. Davis, P.O. Brown, Science 270 (1995) 467.
[20] A.E. Warsen, M.J. Krug, S. LaFrentz, D.R. Stanek, F.J. Loge, D.R. Call,
Appl. Environ. Microbiol. 70 (2004) 4216.
[21] M.J. Hessner, X. Wang, K. Hulse, L. Meyer, Y. Wu, S. Nye, S.W. Guo, S.
Ghosh, Nucl. Acids Res. 31 (2003) e14.
[22] X. Zhou, J. Zhou, Anal. Chem. 76 (2004) 5302.
[23] R.Q. Liang, W. Li, Y. Li, C. Tan, J.X. Li, Y.X. Jin, K.C. Ruan, Nucl. Acids
Res. 33 (2005) e17.
[24] J.S. Shumaker-Parry, M.H. Zareie, R. Aebersold, C.T. Campbell, Anal.
Chem. 76 (2004) 918.
[25] S.J. Chen, Y.D. Su, F.M. Hsiu, C.Y. Tsou, Y.K. Chen, J. Biomed. Opt. 10
(2005) 034005.
[26] E. Rockenbauer, K. Petersen, U. Vogel, L. Bolund, S. Kølvraa, K.V. Nielsen,
B.A. Nexø, Cytometry 64A (2005) 80.
[27] R.J. Fulton, R.L. McDade, P.L. Smith, L.J. Kienker, J.R. Kettman, Clin.
Chem. 43 (1997) 1749.
[28] H. Xu, M.Y. Sha, E.Y. Wong, J. Uphoff, Y. Xu, J.A. Treadway, A. Truong,
E. O’Brien, S. Asquith, M. Stubbins, N.K. Spurr, E.H. Lai, W. Mahoney,
Nucl. Acids Res. 31 (2003) e43.
[29] X. Gao, S. Nie, Anal. Chem. 76 (2004) 2406.
[30] L. Wang, W. Tan, Nano Lett. 6 (2006) 84.
[31] A.R. Clapp, I.L. Medintz, H.T. Uyeda, B.R. Fisher, E.R. Goldman, M.G.
Bawendi, H. Mattoussi, J. Am. Chem. Soc. 127 (2005) 18212.
[32] J. Aldana, Y.A. Wang, X. Peng, J. Am. Chem. Soc. 123 (2001) 8844.
[33] H. Matoussi, J.M. Mauro, E.R. Goldman, G.P. Anderson, V.C. Sundar, F.V.
Mikulec, M.G. Bawendi, J. Am. Chem. Soc. 122 (2000) 12142.
[34] H.C. Cheung, Topics in Fluorescence Spectroscopy. Vol. 2: Principles,
Plenum Press, New York, 1991.
[35] R. Sjöback, J. Nygren, M. Kubista, Spectrochim. Acta A 51 (1995) L7.
[36] S. Hohng, T. Ha, ChemPhysChem 6 (2005) 956.
[37] A.R. Clapp, I.L. Medintz, J.M. Mauro, B.R. Fisher, M.G. Bawendi, H.
Mattoussi, J. Am. Chem. Soc. 126 (2004) 301.
[38] Evident Technolgies, 2006. http://www.evidenttech.com/nanomaterials/
evidots/quantum-dot-specifications.php.
[39] X.F. Wang, U.J. Krull, Anal. Chim. Acta 470 (2002) 57.
[40] X.F. Wang, U.J. Krull, Bioorg. Med. Chem. Lett. 15 (2005) 1725.
[41] X.F. Wang, U.J. Krull, J. Mater. Chem. 15 (2005) 2801.